Following a flood of new, cheap and low-quality competitors, the vapour category had become heavily commoditised — and both volume and value were under pressure.1
Following the category down was the easy answer — match the low prices, hold volume, accept the thinning margins. But that road led nowhere good for volume or profit. So we asked a harder question: was the premium space genuinely closed off, or had the category simply agreed to stop competing for it? And if a premium repositioning could work, could we prove it before committing serious resource? Here's how we found out.
A flood of new, cheap and low-quality competitors had dragged the whole category into the value end — limited price variation, bunched at the bottom, with volume and value both under pressure.12 The knock-on effect was just as damaging: with no credible premium tier, premium-minded consumers had nowhere to trade up to, and took their money elsewhere in the portfolio instead.
The idea was to break from the race to the bottom and reposition Vuse Go into a significantly more premium space. Commoditisation wasn't a property of the category so much as a habit every brand had fallen into together — every other consumer category sustains price segmentation, and vapour had simply stopped building it. A credible premium move could reopen the tier the category had abandoned, and give consumers a real reason to trade up.
The AI didn't replace consumer research — it helped define what should go in front of consumers. Far more went into review than any panel could see: a greater breadth of pack designs, messaging and campaign routes, all evaluated against category data and brand truth, then shortlisted. Only the strongest candidates carried forward into more focused research with real people — so the human-research budget concentrated where it mattered instead of being spread thin across a long list.
Following successful consumer research, a second layer of AI extended that work rather than replacing it — synthetic audiences, modelled on the real consumer data, used to extrapolate the findings across very different markets and stress-test global relevance before anything was committed at scale. The AI never made the call; the strategic decision stayed human. Its job was to widen the funnel early and pressure-test it late.
The result was a validated premium proposition — which led to a rollout of more premium packaging and messaging, and a significantly elevated campaign visual identity, with far stronger premium & quality cues. Evidenced before launch rather than after, and ready to activate — work that has since contributed to Vuse extending its global value-share leadership, with top-markets value share up 1.3 points year-to-date.3
Fuller commercial detail — the premium purchase-intent uplift and the underlying category data behind the repositioning — is available on request. Request the full case study →
Case study describes work led by Andy Parton in a senior role at British American Tobacco plc. All brands and trademarks are the property of their respective owners; Parallax Advisory is independent and not affiliated with or endorsed by them. Figures cited are drawn from the public sources noted; where results are commercially confidential they are shown directionally. Outcomes reflect a specific engagement and are not a guarantee of future results.
FAQ
AI consumer research uses trained models to test creative, packaging and message variants at scale, in days rather than months. It's complementary to traditional research, not a replacement — and it works on both sides of the human stage: widening the field beforehand so real consumers see a stronger shortlist, and extending the findings afterwards, using synthetic audiences to pressure-test relevance across markets. Faster and cheaper, with human research concentrated where it matters most.
Yes, but the move has to be backed by validated consumer evidence at every gate. Commoditisation is a collective brand decision, not a structural property of the category. Breaking ranks requires absolute clarity on pack, flavour, format and message before any market spend — which is exactly what gated AI validation gives you.
Traditional consumer research is expensive and slow, which caps how many ideas you can ever put in front of people. AI synthetic testing removes that cap: far more packaging routes, messages and creative combinations can be evaluated than a conventional research programme could accommodate. That breadth is the point — by the time we committed to large-scale consumer research, we were testing a shortlist we were genuinely confident in, rather than hoping a handful of routes happened to be the right ones.
Each AI workstream produces a hypothesis, not a launch. Pack, message and campaign each pass a human senior validation gate before progressing — claims-risk review, regulatory check, brand-strategy fit. The gates are designed in, not bolted on; they slow no one down because the workflow assumes them from step one.
On Vuse Go, the full global brand work was completed inside a tight strategic window — substantially faster than traditional premiumisation programmes that sequence research, design and validation linearly. Speed comes from running the workstreams in parallel and gated, so validation happens alongside development rather than after it.
Through pre-launch purchase-intent and brand-recognition testing against the category norm. On Vuse Go, three design routes were put in front of target consumers and the strongest carried into range on the back of purchase intent and recognition — so the premium bet was evidenced before any market budget was committed, not after.
Some additional detail can be shared under NDA — please request the full case study or start a conversation below.